Faceted Classification Goodness
Some posts by people a lot more cleverder than me on the XFML list: anyone interested in faceted classification and thesauri should read these.
Travis from Facetmap makes some great points:
"(about topics having multiple parents) [...] The example given in ISO2788 is that "organs (musical instruments)" can have the parents "wind instruments" and "keyboard instruments".
[...]
Many readers of this list probably see where I'm going with this. "wind instruments" and "keyboard instruments" don't belong in the same taxonomy. They are separate facets of "organs". "Wind" goes in the "sound production" facet taxonomy along with "percussion", "string", etc. "Keyboard" belongs in "input devices", which would be a
different facet taxonomy because it describes a different aspect of the instrument.
[...]
Having written an XFML implementation from scratch, I would also like to add this: The relational database graph theory involved is much more elegant, and more efficient to run, if you limit facet nodes to a single parent. The practical upshot of such a limitation is, of course, that a facet mapping engine can handle many more facet nodes and resources before it starts to get swamped."
... and gets responses from the library camp:
"For many years the literature on thesaurus construction has been emphasising the need for facet analysis as an underlying tool to ensure that valid thesaurus relationships are created. It has been good practice to present a controlled vocabulary both as a faceted classification and a thesaurus, since Jean Aitchison "Thesaurofacet" was published in 1969.
[...]
I think we may be talking slightly at cross-purposes here. It may be helpful if I set out the definitions of some of the terms used in the library / information science community, as people coming to this problem from an information technology or "topic map" background have adopted different terms for the same things, which leads to some misunderstandings. [...]"